It is a common observation among computists that man has always attempted to explain the processes of his mind in terms of his latest discovery. Once it was gears and levers, then hydraulic models, then telephone switchboards. Now it seems obvious that the mind is an information-processing organ: a computer.
This recent metaphor is one that is likely to last quite a long time, since it refers to a machine with a broad power to employ symbols. Many in the humanities (without knowing precisely why) have been repelled by the image, and so have never paid their admission to enter, and see at first hand, the charms of the idea. Two books have now been published which make it much easier for such non-specialists.
One of the two volumes, "Machines Who Think," by Pamela McCorduck, is both cheaper and better as a general introduction to the puzzle of machines and thinking. As we all know, most computers in the world are still doing nothing but adding, subtracting, and giving clerks a perfect cover for their own mistakes. But McCorduck writes that computers are not by nature dull and inflexible. As some humanistic computists have pointed out, computers have always been perfectly capable of addressing people by their names -- and yes, they can without blinking handle nicknames, name changes and cases of two or more people with the same name. We are numbers in accounts and files because people wanted it that way.
It is only the avant-garde of computer science, in a little domain called "artificial intelligence," that we find the computer in its natural state. McCorduck's book is a casual history of that field, and it is without doubt the best book of the subject so far published for the general reader. Sadly, its appearance does the new science no good -- passages in unintelligible dialects of computerese are sprinkled throughout the book, and footnotes, caps and brackets grow everywhere like weeds. However, there are also lively anecdotes and brief personality sketches of the prominent researchers.
McCorduck begins with the question implied in her title. "'Can a machine think?' . . . Does a chicken have lips? . . . Yet for all its absurdity we find the idea irresistible. Our history is full of attempts -- nutty, eerie, comical, earnest, legendary, and real -- to make artificial intelligences, to reproduce what is the essential us."
From before the time of Homer, mechanical statues have bowed, shook and glided to strike the clock, pour the water, open the door. And with each age the devices have become more elaborate and the mathematics of the time more ready to accommodate the present machine -- the computer -- which, at least in principle, is capable of intelligent action. McCorduck confronts directly the question of fear. "We approach the alien [machine intelligence] with deeply mixed feelings, part terror and part exhilaration," she writes.
Then she suggests a passage from Joseph Conrad and neatly describes her feeling, the passage in which Marlow, traveling into the heart of darkness, comes face-to-face with native Africans in their home surroundings, and he is shocked by what he finds: "We are accustomed to look upon the shackled form of a conquered monster, but there you could look at a thing monstrous and free. It was unearthly, and the men were -- no, they were not inhuman. Well, you know, that was the worst of it -- the suspicion of their not being inhuman . . . what thrilled you was just the thought of their humanity -- like yours -- the thought of your remote kinship, with this wild and passionate uproar."
And McCorduck adds to this, "When we come face to face with the idea of thinking machines we have much the same reaction. What thrills us -- in the deepest sense -- is the thought of our remote kinship with these contrivances." The fear of computers was so widespread in the 1950s, McCorduck says, that IBM itself took a metaphorical hatchet to the works: "Sales executives at IBM began to grow nervous lest the very machines they were trying to sell prove so psychologically threatening that customers would refuse to buy them. Thus they made a deliberate decision to defuse the potency of (intelligent) programs by conducting a hard-sell campaign picturing the computer as nothing more than a quick moron." So computers have continued to shuffle about doing the menial mental jobs.
In "The Computer Age," we listen to the voices of the scientists themselves. We leave behind us the stories and faces of the researchers shining through the pages, and we see the issues raw: over the next 20 years what will computers do to us in economics, in culture, in education, in business, in science, in the home?
There are 22 essays in this fat book, but it's just a few of them that give the book the greatest part of its value. Three of the best are Terry Winograd's piece on "convivial computing," Roger Noll's work on how government regulation may affect public computing services in the future, and, the best of all, Marvin Minsky's brilliantly clear primer on intelligence and the difficulties of creating its artificially. c
While McCorduck's book provides the general and historical picture of computing. "The Computer Age" gives more detail on a broad range of computer issues. Together, the books make a fine pair for the general reader. c